The Meta dilemma: Invest billions in AI but find ways to cut emissions too
The social networking giant sees AI as important, but also wants to use it for energy and water management, materials design. Read More
Meta will spend $38 billion-$40 billion in capital expenses in 2024, roughly a quarter’s worth of revenue. Those infrastructure investments will accelerate in 2025 for the social networking company — thanks mostly to rise of its use of artificial intelligence, Meta said in its third-quarter 2024 earnings report.
Meta is using AI to improve video recommendations and other content shown in users’ information feeds, which has increased time spent on Facebook and Instagram by 8 percent and 6 percent, respectively. More than 1 million advertisers are using its generative AI creation tools, and the Meta AI assistant now has more than 500 million monthly active users.
“It’s clear that there are a lot of new opportunities to use new AI advances to accelerate our core business that should have strong ROI over the next few years,” said Meta CEO Mark Zuckerberg during Meta’s Oct. 30 call with analysts. “So, I think we should invest more there. Second, our AI investments continue to require serious infrastructure, and I expect to continue investing significantly there too.”
Data center growth at odds with emissions cuts
New data centers being built at a furious pace by big digital apps companies such as Meta and the Big 3 cloud service providers Amazon, Google and Microsoft are straining the electric grid’s capacity. That growth is also at odds with the companies’ emissions reduction goals.
Meta emitted 14,067,104 metric tons of carbon dioxide equivalent in 2023, up slightly from 2022 and more than double what it reported in 2019, according to Meta’s 2024 sustainability report. The company’s goal is to cut Scope 1 and 2 emissions by 42 percent in 2031, compared with its 2021 baseline year. Instead, Meta’s carbon footprint is up about 38 percent.
“Obviously, the load growth that we’re experiencing, the amount of energy that’s being built and invested in, makes it more challenging on sustainability goals,” said Blair Swedeen, global head of net zero and sustainability at Meta, during an Oct. 30 session about AI at the VERGE 24 conference. “We’re 100 percent committed on our sustainability goals, our public commitments that we’ve made, and so it’s even more important to lean in on the innovation piece.”
How AI aids Meta’s climate goals
Like a growing number of other companies, Meta’s sustainability team uses AI to gather data for ESG disclosures. It also uses AI to identify opportunities to reduce greenhouse gas emissions or improve energy efficiency, Swedeen said.
“It’s about being able to run all the simulations, process all of that data, and then come up with judgments or recommendations,” he said.
Two examples Swedeen shared:
- A model developed by Meta and the University of Illinois that assesses concrete mixtures for data center construction for strength and emissions properties. By substituting materials such as fly ash, slag or ground glass, the approach helps reduce the footprint of the material by about 40 percent. Doing this manually is time-consuming.
- In another project, Meta used AI to adjust the way that fans work for cooling hardware in the data center. Meta’s team ran simulations using AI and was able to reduce energy consumption related to the fans by 20 percent.
AI is fundamental to ongoing efficiency initiatives across Meta’s data center infrastructure and for driving out both costs and carbon emissions. “It’s just this constant process of iterating,” Swedeen said.
[Learn what’s next in decarbonization, disclosure, supply chains and more at GreenBiz 25 — our premier sustainability event, February 10-12, Phoenix, Arizona.]
Subscribe to Trellis Briefing
